Ashdin Publishing

Journal of Evolutionary Medicine

Vol. 2 (2014), Article ID 235678, 8 pages
doi:10.4303/jem/235678

ASHDIN

publishing

Review Article
THEORY

Using Systems Biology to Understand Cancer as an Evolutionary Process

John W. Pepper, Barbara K. Dunn, Richard M. Fagerstrom, John K. Gohagan, and Nadarajen A. Vydelingum
Division of Cancer Prevention, National Cancer Institute, 6130 Executive Blvd. MSC 7315, Bethesda, MD 20892-7354, USA
Address correspondence to John W. Pepper, pepperjw@mail.nih.gov
Received 17 December 2012; Revised 17 December 2013; Accepted 18 December 2013
Copyright 2014 John W. Pepper et al. This is an open access article distributed under the terms of the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract Unsatisfactory progress in cancer medicine and prevention

calls for new research approaches. Research can broaden its view
of cancer to include not only specific molecular elements, but also
the process that explains their origin and dynamics. This process is
Darwinian evolution of somatic cells. Applicable modeling techniques
are available from process-oriented systems biology. We review
relevant concepts and techniques, and their application to four key
open questions in cancer prevention research. Helpful concepts are
transferable from classical evolutionary biology and ecology, while
useful techniques include computational agent-based modeling. The
research questions we review include (1) why do benign neoplasms
often progress to malignancy? (2) what is the chronological sequence
of molecular events in cancer progression? (3) how can we find
reliable molecular biomarkers for cancer? and (4) will evolved drug
resistance stymie efforts at a long-term cancer chemoprevention? We
conclude that molecular analysis can be usefully augmented with
process-oriented systems biology to guide empirical research into the
most productive directions.
Keywords systems biology; cellular evolution; clonal evolution; cancer prevention

1. Introduction
Despite the US war on cancer declared in 1971, cancer
incidence and mortality have fallen less than initially hoped
and expected [7]. Even within the most recent five years,
cancer incidence in the US declined less than 1%, and
cancer mortality declined less than 2% [64]. This has led
some researchers to argue that new approaches to cancer
medicine are needed, including a conceptual framework that
can explicitly address the complex dynamics that arise from
the cellular evolution and adaptation that occurs in cancer
as it interacts with its own microecology [18, 26].
Compared to the areas in which medical research had
its most dramatic successes, cancer presents fundamentally
different challenges, because it arises through a process
that is endogenous to human tissues. The research strategy
of using molecular reductionism to define a fixed diseasecausing entity has a brilliant track record in the fields of
infectious disease and inherited genetic disease, but has
been less successful in cancer biology. Molecular analysis

quickly revealed the consistent genetic nature of nonhuman

pathogen species, such as bacteria, and of inherited genetic
defects. In contrast, cancer reflects a complex process of
ongoing genetic change in human somatic cells, and it usually has no fixed molecular basis. Thus, in important ways,
a cancer is not simply a thing to be removed or destroyed,
but also a process to be prevented, controlled or managed.
Although cancer risk can be greatly increased by inherited genetic factors or infectious pathogens, neither of those
is necessary or sufficient on its own to be the cause of most
cancer. Because cancer arises from a process that is endogenous to human tissues, it is often independent of inherited
genes or external insults. Cancer biology can most naturally be organized around an understanding of this endogenous process of somatic cellular evolution, which consists
of mutations causing differential reproduction, with inheritance, among somatic cells. Somatic evolution is a special
case of the Darwinian process of evolution by natural selection. Process-oriented approaches from systems biology can
help to encompass the complexity of somatic cellular evolution, making cancer more comprehensible and tractable.
2. The central process of cancer biology: somatic cellular
evolution
As early as the 1970s, Cairns [12] and Nowell [50]
proposed that somatic cell mutation, selection, and resulting
Darwinian evolution are the fundamental processes by
which neoplasms arise, acquire malignancy, and evade
therapy. This hypothesis lay dormant for decades before
being further developed and confirmed with empirical
results from modern cancer biology [14, 21, 29, 56, 57, 67].
Currently though, cancer development at the cellular level
is widely regarded as a Darwinian evolutionary process
involving natural selection of genetically variant cells in
the context of a complex microenvironmental ecology [6].
Our only amendment to the foregoing quote would be to

Journal of Evolutionary Medicine

explicitly recognize that relevant mechanisms of cellular

heredity include both genetic and epigenetic variation [57].
In common terminology, cellular selection and evolution are equivalently referred to as clonal selection and
evolution, because each dividing somatic cell produces a
clone of progeny with a shared genome and a shared fate.
As molecular techniques have provided an increasing
wealth of data, the hypothesis of cellular evolution in cancer
has been supported in multiple cancer types. For example,
cellular evolution as a basis for neoplastic progression and
changes in cancer has been documented in molecular detail
in breast cancer [47, 62, 61], pancreatic cancer [13], renal
cancer [28], ovarian cancer [48], leukemia [6, 70], colorectal
cancer [32], and head and neck cancer [42].
The hypotheses and observations of cellular evolution
have led to an emerging scientific theory of cancer
dynamics, which elucidates origination, progression,
metastasis, and response to treatment. This development
is greatly assisted by a mature literature of concepts and
techniques from the classical study of Darwinian evolution
in populations of organisms. One optimistic assessment is
that, despite the extreme complexity of cancer, most, if
not all, of this complexity can be explained by classical
evolutionary principles [29].
3. Computational
processes

tools

for

studying

evolutionary

Armed with an understanding of the basic process behind

the complex dynamics of cancer, researchers are increasingly adopting systems biology approaches that have
proven effective for understanding complex dynamic
processes such as cellular evolution. These approaches allow
researchers to reveal not only the molecular components of
cancer cells, but also the reasons why molecular components
vary among cells, and why they change over time during
cancer progression. Cancer is arguably the ultimate complex
biological system [31]. For coping with such complexity,
systems biology approaches have proven effective in much
of biomedicine, including cancer biology [8, 44, 60], and
these approaches fall into two distinct categories. One consists of techniques for analyzing large omics data sets [1,
33]. The other category consists of mathematical and computational models for representing underlying processes, as
opposed to a specific set of resulting data [4, 5, 31]. Here, we
refer to this second category as process-oriented systems
biology. Without using that label for it, other authors have
suggested that process-oriented systems biology has the
potential to play a central role in cancer prevention [27].
One specific technique for conducting process-oriented
systems biology is the use of agent-based computer
models [39]. In these models, each cell can be explicitly
represented by a computational agent (a piece of
software) that represents a cell as it interacts with its

microenvironment, mutates, divides, passes on acquired

somatic mutations, and thereby founds a new clone. In
this technique (also called bottom-up simulation, or
microsimulation), the traits and behaviors of cells are
specified in advance, but larger scale patterns and processes
are not. Only as the model is used to run a simulation, often
with a large and heterogeneous population of virtual cells,
do larger scale patterns and processes emerge, often with
collective outcomes that were not intuitively obvious in
advance. This can generate a process of digital evolution
inside a computer, which can provide a detailed, transparent,
and predictive model of cellular evolution and progression
under a specified set of conditions [2, 52]. The results of such
simulations are not empirical data, but they can generate
incisive new hypotheses that are empirically testable, and
that can guide empirical research into productive new
directions. Far from substituting for empirical research,
computer models are most effective when they are closely
integrated with experimentation, and are used to refine
hypotheses and better focus experiments [17, 22].
Other techniques are also useful for modeling in
process-oriented systems biology. For example, the
authors in [9], among others, have represented evolutionary dynamics using a game theoretic framework for
mathematical analysis. Their results elucidated the role of
tissue interactions in directing the progression of prostate
cancer into distinct courses, ranging from indolent to highly
life-threatening.
4. Process-oriented systems biology in cancer prevention
research
The process of somatic cellular evolution can rarely
be observed directly, and most empirical data can only
represent a snapshot in time from this ongoing process.
Explicitly defining the process thought to underlie such
observable data extends their explanatory power. Below,
we consider four current questions in cancer prevention
research, and discuss a process-oriented approach to each
of them.
4.1. Why do benign neoplasms often progress to malignancy?
The defining (and deadly) feature of cancer, as opposed to
benign growths, is local tissue invasion, ultimately leading
to distant metastasis. Because this process emerges from
interactions among multiple cells and their microenvironments, process-oriented system models are particularly well
suited to its study. For this purpose, tools have been borrowed from both classical ecology and evolutionary biology.
Ecologists have used computational studies of spatiotemporal dynamics to compare tissue invasion by cancer
cells with the invasion of natural ecosystems by artificially
introduced species [40]. These authors posit that their results

Journal of Evolutionary Medicine

can help explain the patterns of genetic diversity observed

both within and among tumors in a patient. They also
describe a distinct geometrical signature of invasion
dynamics in both ecological landscapes and tumors,
which they suggest may someday be used along with
advanced imaging technologies to accurately discriminate
malignancies from benign growths [41]. Independently,
biomedical researchers have also found that computer
simulations of mathematical models closely resemble the
actual morphologies and spatial patterns of tumor growth
and invasion, and that predictions of the model are met
by observations of real tumors in vivo [23]. This group
agreed with the ecologists cited above that results from
process-oriented models can make tumor invasion more
recognizable, and they further predicted that different
invasion morphologies would correspond to different stages
of tumor progression [10]. More recently, agent-based
models have even been calibrated to match the precancerous
breast tumors of specific patients, to better track and predict
their progression, as observed by mammography [38].
Beyond mere prediction, a central goal of cancer
prevention is to block progression from benign premalignancies to cancer. At a cellular level, this progression
entails the acquisition of motility by formerly sedentary
neoplastic cells. In seeking ways to block this change, it
will help to understand what causes it. Building on classical
evolutionary ecology, one research team [3] hypothesized
that the intense local competition for resources caused by
abnormally rapid cell proliferation in neoplasms selects for
motile cells that can thrive by leaving the resource-depleted
zones where other neoplastic cells languish and die, and that
this selection drives the evolution of somatic cell motility.
Agent-based computational models provided support for
this hypothesis. Their results also suggest that therapeutic
agents designed to block cell motility, instead of killing
cancer cells, will be more robust against acquired drug
resistance, and while blocking malignancy, will also select
for reduced cell proliferation and tumor growth [3].
4.2. What is the chronological sequence of molecular events
in cancer progression?
A long-standing challenge in cancer biology is to understand
and predict the chronological sequence of molecular events
underlying the initiation and progression of tumors [19].
Many early genetic models assumed that tumors were
genetically homogenous, and thus could be characterized
as having a single tumor genotype. On this basis, the
classic type of model was a linear sequence of events, or a
canonical path model, constructed from cross-sectional
data on multiple patients [65]. However, more recent
evidence revealed so much genetic heterogeneity within
tumors that the assumption of one single genotype per
tumor is clearly no longer valid [49, 63].

Agent-based evolutionary models of tumor progression

suggest that the temporal, or evolutionary, order of mutations acquired in the clones that survive over the lifetime of a
neoplasm may not be consistent with the path order inferred
from cross-sectional data [65]. Because cellular evolution
during tumorigenesis is a unique process in each patient, it is
not valid to assume that the state of one tumor is informative
for the history of even an apparently similar tumor in a different patient. Independent evolutionary trajectories in each
patient make it difficult to reconstruct temporal order from
cross-sectional data. In contrast, observed cellular diversity
can be analyzed with phylogenetic methods from classical
evolutionary biology to reconstruct cell lineages within individual tumors. This can accurately reveal the true temporal
order of events for the clones within a specific tumor (Figure 1).
The two key results of this study [65] are (1) that crosssectional data can be misleading because of independent cellular evolution within each patient and (2) that even though a
detailed study of intratumor variation can reveal the average
ordering of molecular events within that tumor, not all cells
in the tumor actually reflect that average order of events.
As with any results from rigorous simulations, these should
not be treated as final conclusions, but rather as plausible
hypotheses and strong candidates for empirical testing [34].
4.3. How can we find reliable molecular biomarkers for cancer?
Another major research direction in cancer prevention is
the development of cancer biomarkers for early detection
and for prognosis. To date, most biomarker development
has not been informed by evolutionary dynamics, but
instead has focused on specific molecules. This strategy has
identified many candidate somatic genetic and epigenetic
biomarkers of cancer or cancer risk. However, very few
specific marker molecules have proven to be reproducibly
effective for identifying cancer [59]. Similarly, no specific
molecule has yet been validated as a robust predictor of
progression to cancer, or been useful to reduce cancer
mortality [35].
One likely reason for this lack of success is that the
process of cellular evolution is fueled by stochastic mutation, which is unpredictable and varies among individuals.
Because of shared selective pressures for survival and proliferation, different cancer cases tend to converge on the same
hallmark cell traits [30]. However, because somatic mutation is stochastic, the molecular basis for these traits can
differ, and at a molecular level, each cancer can be unique
in many ways [29], making it difficult to find molecular
biomarkers that will be consistent for different patients.
Instead of seeking consistency across individuals
in cancer-specific molecules, a different approach is to
recognize that the cellular genetic diversity of neoplasms

Journal of Evolutionary Medicine

Figure 1: Reprinted with permission from Sprouffske et al. [65]. The temporal order of mutations seen in simulated cancer
clones rarely matches the path order inferred from cross-sectional data, but it does match the order inferred from genetic
dependency analysis of intratumor data. (a) Plotting the percentage of tumors with a given mutation at increasing neoplasm
sizes can be used to infer. (b) The cross-sectional path model of mutations. (c) However, the proportion of cells within a
simulated neoplasm that has a history consistent with the inferred cross-sectional path order tends to be low (mean [SEM]
= 7.3% [1.0%], n = 90 simulation runs). (d) In contrast, the proportion of cells within a simulated neoplasm that has a
history consistent with order inferred from the genetic dependency analysis is high (mean [SEM] = 99.7% [0.1%], n = 90).
Each mutation type is represented by a different color in panels (a) and (b): loss of differentiation (LD) is green, evasion
of apoptosis (EA) is light purple, limitless replicative potential (LR) is orange, sustained angiogenesis (SA) is dark purple,
genomic instability (GI) is light green, self-sufficiency in growth signals (SG) is light blue, and insensitivity to antigrowth
signals (IA) is dark blue.
fuels their evolution, and to measure this diversity itself as
a marker, instead of any single molecule. Classical evolutionary theory states that the rate of evolutionary change is
proportional both to trait variance and to fitness variance in
the members of the evolving population [58]. Thus greater
genetic variation among cells in a neoplasm is expected
to drive faster progression. There is substantial empirical
evidence supporting this hypothesis. In one study [71], five
different cell lines representing various types of cancers
were analyzed for karyotypic diversity among cells. In each
of these lines, the highest level of chromosomal diversity
was coupled with the strongest tumorigenicity (Figure 2).
This pattern is not limited to karyotype, or to in vitro cell
lines. High cellular diversity accurately predicted risk in
patients monitored for progression from Barretts esophagus
to esophageal cancer, and simultaneously monitored by
whole-genome sequencing of repeated biopsies. Cell
diversity was measured in SNPs, copy number, loss of

heterozygosity, and aneuploidy [36]. Every tested measure

of genetic diversity among cells produced highly significant
(P < .001) predictors of progression to cancer [43]. Similar
patterns have been reported in other human tumors. In a
study of human breast carcinomas, an index of cellular
genetic diversity based on copy number ratios was higher
in invasive than in situ components from the same section
of the same tumor [51]. Cellular genetic diversity was also
positively correlated with tumor grade [51, supplemental
Table 7]. (Note that these authors considered this latter trend
to be only suggestive, needing confirmation in a larger
sample set.)
Taken together, these experimental results and observational clinical results suggest that the genetic diversity
that fuels somatic cellular evolution may be a robust
general biomarker of cancer risk and prognosis. Such a
diversity marker, if proven in translational research, could
simultaneously help to address two major challenges:

Journal of Evolutionary Medicine

Figure 2: Reprinted with permission from Ye et al. [71].

Comparison of the spectral karyotypes of two cell
sublines, each derived in vitro from the same cell line
(MCF10A3, from a human breast carcinoma in situ).
The high-diversity subline CSC-3 (panel (b)) produced
tumors in immunodeficient mice, while the low-diversity
subline CSC-1 (panel (a)) did not. Only subline CSC-3
with xenograft tumorigenicity (panel (b)) contained many
low-frequency chromosome aberrations not found in both
sublines (indicated by the yellow shaded boxes), in addition
to the five high-frequency chromosome aberrations found
in both (indicated by blue-shaded boxes).
achieving early detection of truly dangerous neoplasms, and
avoiding overdiagnosis of nonthreatening growths.
4.4. Will evolved drug resistance stymie efforts at long-term
cancer chemoprevention?
Developing drugs for chemoprevention is a major line of
research in cancer prevention that may face pitfalls from
cancer evolution. These potential pitfalls have already been
demonstrated in cancer therapy. Many agents have been
developed and successfully used for selectively killing
cancer cells. A very general limitation on this approach,
however, is the acquired drug resistance that results from
cellular evolution [54]. This makes resistant relapse a
routine sequel to initial tumor shrinkage during therapy [15,
54], and most patients who die of cancer are killed by a
cancer that has evolved drug resistance. The problem of
acquired resistance may prove to be even more limiting for
a long-term chemoprevention than it is for therapy. While
successful therapy is completed within a shorter time frame,
cancer prevention might ideally continue throughout life.
Thus, any drugs that can be administered effectively only
for a limited time may not be entirely satisfactory.

There is clinical evidence that the time limits on

effective drug administration seen during cancer therapy
can also arise in chemoprevention. This is illustrated by the
selective estrogen receptor modulators (SERMs) tamoxifen
and raloxifene. Both of these have recently been tested
and approved as chemoprevention agents for breast cancer
in high-risk women [24, 68, 69]. However, the problem
of acquired drug resistance in chemoprevention has not
yet been adequately addressed. In cancer therapy, many
initially responsive tumors later develop acquired resistance
to tamoxifen [45, 46]. A similar acquired resistance to
raloxifene, as well as cross-resistance between these two
SERMs, has been demonstrated in vitro with breast
carcinoma cells [25, 37, 66]. This established pattern
of acquired resistance to SERMs appears to arise in
chemoprevention as well as in chemotherapy. Controlled
chemoprevention trials showed a reduction in breast cancer
incidence in healthy high-risk women after using tamoxifen
for five years [20]. However, in the adjuvant treatment
setting, longer administration does not confer further benefit
in preventing new primary tumors [20]. This result suggests
that long-term intervention to prevent a new cancer may
suffer from the same limitations as long-term chemotherapy
to prevent recurrence, and for the same reason of cellular
evolution of acquired drug resistance. Achieving life-long
protection through chemoprevention may require longer
periods of effective drug administration, and thus may not
be possible with any drug that selects for drug-resistant
variants. Whenever somatic cells are dividing and subject to
mutation, resistant variants are likely to arise. Any drug that
kills or suppresses individual drug-sensitive cells thereby
selects more resistant cells to survive with reduced competition after the removal of drug-sensitive neighbors [53].
This leads to a rapid out-growth, or clonal expansion, of
a lineage of resistant cells which quickly takes the place of
the drug-sensitive cells removed by treatment [54].
Providing the best possible protection through chemoprevention may require ways to delay cellular evolution of
acquired resistance. Application of evolutionary theory
from other contexts suggests specific strategies [53].
Therapeutics that change the tumor microenvironment
to make it less hospitable to tumor growth and invasion can
impede all cancer cells equally, without selecting among
them and thereby driving the evolution of resistance. For
example, nonsteroidal anti-inflammatory drugs meet this
criterion, and there is some evidence that they may remain
effective indefinitely for long-term chemoprevention [16].
Other potential drug classes could also exert less
selection on existing variation, and thus slow the evolution
of acquired drug resistance. Potential new drugs could
exploit the fact that cancer cells construct their own
microenvironment by producing, and depending on, many
shared substances that increase the capacity of their shared

Journal of Evolutionary Medicine

initially normal human cells. This process is best understood

as somatic cellular evolution. Process-oriented systems
biology provides tools for modeling this process to generate
the most crucial hypotheses for empirical testing.
Acknowledgments For helpful input, the authors thank Carlo C.
Maley and Brian J. Reid.

References

Figure 3: Simulation results reprinted from a previous

work [55]. Frequency of drug-resistant cells after 1,000
cell generations of mutation and evolution. High transfer
coefficients correspond to highly diffusible drug targets
produced by one cancer cell that provide benefits to
other nearby cancer cells. A transfer coefficient of zero
corresponds to a cell-intrinsic target molecule that is not
shared among neighboring cells. Each marker represents the
mean for a different drug dosage: filled circle = 100% of full
dose, open circle = 75%, triangle = 50%. Error bars show
standard error across 10 simulation runs using different seed
values for the pseudorandom number generator.
microenvironment to support tumor growth and invasion.
Examples include secreted angiogenesis factors, growth
and invasion factors, and immune suppression factors [53,
54]. In agent-based computer simulations, drugs targeting
such shared public goods products do not single out the
specific cells producing them, or those cells producing drugresistant variants of them. Thus they do not select strongly
for resistant cells, and do not strongly drive the evolution
of acquired resistance [55] (Figure 3). Thus such drugs
should retain longer-term effectiveness both in treatment,
and also in chemoprevention. Specific targets for potential
drugs of this class have been proposed, and in some cases
successfully tested in animal models [53, 54]. The theoretical prediction that targeting the tumor microenvironment
will produce less acquired drug resistance has also been
supported clinically for some antiangiogenic drugs [11].
5. Conclusions
Viewing cancer simply as an entity to be removed or
destroyed can overlook important aspects of cancer biology,
including why cancer is dynamic and adaptive, and why
it is difficult to target as a nonself entity. Molecular
reductionism contributes greatly to advances in cancer
research, but is even more useful within the framework of
understanding the somatic cellular evolution that generates
the diverse molecules observed in cancer cells. At the heart
of cancer biology is a change over time in populations of

[1] M. S. Abu-Asab, M. Chaouchi, S. Alesci, S. Galli, M. Laassri,

A. K. Cheema, et al., Biomarkers in the age of omics: time for a
systems biology approach, OMICS, 15 (2011), 105112.
[2] C. Adami and C. O. Wilke, Experiments in digital evolution
(editors introduction to the special issue), Artif Life, 10 (2004),
117122.
[3] C. A. Aktipis, C. C. Maley, and J. W. Pepper, Dispersal evolution
in neoplasms: the role of disregulated metabolism in the evolution
of cell motility, Cancer Prev Res (Phila), 5 (2012), 266275.
[4] G. An, Translational systems biology using an agent-based
approach for dynamic knowledge representation: An evolutionary paradigm for biomedical research, Wound Repair Regen, 18
(2010), 812.
[5] G. An, Q. Mi, J. Dutta-Moscato, and Y. Vodovotz, Agent-based
models in translational systems biology, Wiley Interdiscip Rev
Syst Biol Med, 1 (2009), 159171.
[6] K. Anderson, C. Lutz, F. W. van Delft, C. M. Bateman, Y. Guo,
S. M. Colman, et al., Genetic variegation of clonal architecture
and propagating cells in leukaemia, Nature, 469 (2011), 356
361.
[7] J. C. Bailar 3rd and E. M. Smith, Progress against cancer?, N
Engl J Med, 314 (1986), 12261232.
[8] S. G. Baker and B. S. Kramer, Systems biology and cancer:
Promises and perils, Prog Biophys Mol Biol, 106 (2011), 410
413.
[9] D. Basanta, J. G. Scott, M. N. Fishman, G. Ayala, S. W.
Hayward, and A. R. Anderson, Investigating prostate cancer
tumour-stroma interactions: clinical and biological insights from
an evolutionary game, Br J Cancer, 106 (2012), 174181.
[10] E. L. Bearer, J. S. Lowengrub, H. B. Frieboes, Y. L. Chuang,
F. Jin, S. M. Wise, et al., Multiparameter computational modeling
of tumor invasion, Cancer Res, 69 (2009), 44934501.
[11] T. Boehm, J. Folkman, T. Browder, and M. S. OReilly,
Antiangiogenic therapy of experimental cancer does not induce
acquired drug resistance, Nature, 390 (1997), 404407.
[12] J. Cairns, Mutation selection and the natural history of cancer,
Nature, 255 (1975), 197200.
[13] P. J. Campbell, S. Yachida, L. J. Mudie, P. J. Stephens, E. D.
Pleasance, L. A. Stebbings, et al., The patterns and dynamics of
genomic instability in metastatic pancreatic cancer, Nature, 467
(2010), 11091113.
[14] B. Crespi and K. Summers, Evolutionary biology of cancer,
Trends Ecol Evol, 20 (2005), 545552.
[15] J. J. Cunningham, R. A. Gatenby, and J. S. Brown, Evolutionary
dynamics in cancer therapy, Mol Pharm, 8 (2011), 20942100.
[16] C. Curiel-Lewandrowski, T. Nijsten, M. L. Gomez, L. M.
Hollestein, M. B. Atkins, and R. S. Stern, Long-term use
of nonsteroidal anti-inflammatory drugs decreases the risk of
cutaneous melanoma: results of a United States case-control
study, J Invest Dermatol, 131 (2011), 14601468.
[17] T. S. Deisboeck, Z. Wang, P. Macklin, and V. Cristini, Multiscale
cancer modeling, Annu Rev Biomed Eng, 13 (2011), 127155.
[18] N. Drake, Forty years on from Nixons war, cancer research
evolves, Nat Med, 17 (2011), 757.
[19] E. R. Fearon and B. Vogelstein, A genetic model for colorectal
tumorigenesis, Cell, 61 (1990), 759767.

Journal of Evolutionary Medicine

[20] B. Fisher, J. Dignam, J. Bryant, and N. Wolmark, Five versus
more than five years of tamoxifen for lymph node-negative breast
cancer: Updated findings from the National Surgical Adjuvant
Breast and Bowel Project B-14 randomized trial, J Natl Cancer
Inst, 93 (2001), 684690.
[21] S. A. Frank and M. A. Nowak, Problems of somatic mutation and
cancer, Bioessays, 26 (2004), 291299.
[22] H. B. Frieboes, B. R. Smith, Y. L. Chuang, K. Ito, A. M.
Roettgers, S. S. Gambhir, et al., An integrated computational/
experimental model of lymphoma growth, PLoS Comput Biol, 9
(2013), e1003008.
[23] H. B. Frieboes, X. Zheng, C. H. Sun, B. Tromberg, R. Gatenby,
and V. Cristini, An integrated computational/experimental model
of tumor invasion, Cancer Res, 66 (2006), 15971604.
[24] E. M. Gabriel and I. Jatoi, Breast cancer chemoprevention,
Expert Rev Anticancer Ther, 12 (2012), 223228.
[25] Z. O. Gao, Z. P. Gao, J. Z. Fields, and B. M. Boman, Development
of cross-resistance to tamoxifen in raloxifene-treated breast
carcinoma cells, Anticancer Res, 22 (2002), 13791383.
[26] R. A. Gatenby, A change of strategy in the war on cancer, Nature,
459 (2009), 508509.
[27] R. A. Gatenby, R. J. Gillies, and J. S. Brown, Evolutionary
dynamics of cancer prevention, Nat Rev Cancer, 10 (2010), 526
527.
[28] M. Gerlinger, A. J. Rowan, S. Horswell, J. Larkin, D. Endesfelder, E. Gronroos, et al., Intratumor heterogeneity and
branched evolution revealed by multiregion sequencing, N Engl
J Med, 366 (2012), 883892.
[29] M. Greaves and C. C. Maley, Clonal evolution in cancer, Nature,
481 (2012), 306313.
[30] D. Hanahan and R. A. Weinberg, Hallmarks of cancer: the next
generation, Cell, 144 (2011), 646674.
[31] H. Hatzikirou, A. Chauviere, A. L. Bauer, A. Leier, M. T.
Lewis, P. Macklin, et al., Integrative physical oncology, Wiley
Interdiscip Rev Syst Biol Med, 4 (2012), 114.
[32] D. Hirsch, J. Camps, S. Varma, R. Kemmerling, M. Stapleton,
T. Ried, et al., A new whole genome amplification method
for studying clonal evolution patterns in malignant colorectal
polyps, Genes Chromosomes Cancer, 51 (2012), 490500.
[33] E. Kolker, E. Stewart, and V. Ozdemir, Opportunities and
challenges for the life sciences community, OMICS, 16 (2012),
138147.
[34] D. C. Krakauer, J. P. Collins, D. Erwin, J. C. Flack, W. Fontana,
M. D. Laubichler, et al., The challenges and scope of theoretical
biology, J Theor Biol, 276 (2011), 269276.
[35] X. Li, P. L. Blount, T. L. Vaughan, and B. J. Reid, Application
of biomarkers in cancer risk management: evaluation from
stochastic clonal evolutionary and dynamic system optimization
points of view, PLoS Comput Biol, 7 (2011), e1001087.
[36] X. Li, P. C. Galipeau, C. A. Sanchez, P. L. Blount, C. C.
Maley, J. Arnaudo, et al., Single nucleotide polymorphism-based
genome-wide chromosome copy change, loss of heterozygosity,
and aneuploidy in Barretts esophagus neoplastic progression,
Cancer Prev Res (Phila), 1 (2008), 413423.
[37] H. Liu, E. S. Lee, B. Chen, K. P. Mckian, A. De Los Reyes, and
V. C. Jordan, Cross-resistance of tamoxifen and raloxifene on the
growth of raloxifene-resistant breast cancer cells in vivo and in
vitro, Breast Cancer Res Treat, 69 (2001), 255.
[38] P. Macklin, M. E. Edgerton, A. M. Thompson, and V. Cristini,
Patient-calibrated agent-based modelling of ductal carcinoma
in situ (DCIS): from microscopic measurements to macroscopic
predictions of clinical progression, J Theor Biol, 301 (2012),
122140.
[39] P. P. Maglio and P. L. Mabry, Agent-based models and systems
science approaches to public health, Am J Prev Med, 40 (2011),
392394.

7
[40] D. E. Marco, S. A. Cannas, M. A. Montemurro, B. Hu, and
S. Y. Cheng, Comparable ecological dynamics underlie early
cancer invasion and species dispersal, involving self-organizing
processes, J Theor Biol, 256 (2009), 6575.
[41] M. Martin, Comparing invasive species to metastatic cancers
inspires new insights for modelers, J Natl Cancer Inst, 100
(2008), 8889.
[42] M. Masuda, S. Toh, T. Wakasaki, M. Suzui, and A. K.
Joe, Somatic evolution of head and neck cancer biological
robustness and latent vulnerability, Mol Oncol, 7 (2013), 1428.
[43] L. Merlo, N. Shah, X. Li, P. Blount, T. Vaughan, B. Reid,
et al., A comprehensive survey of clonal diversity measures in
Barretts esophagus as biomarkers of progression to esophageal
adenocarcinoma, Cancer Prev Res (Phila), 3 (2010), 13881397.
[44] M. Mijuskovic, S. M. Brown, Z. Tang, C. R. Lindsay, E. Efstathiadis, L. Deriano, et al., A streamlined method for detecting
structural variants in cancer genomes by short read paired-end
sequencing, PLoS One, 7 (2012), e48314.
[45] E. A. Musgrove and R. L. Sutherland, Biological determinants of
endocrine resistance in breast cancer, Nat Rev Cancer, 9 (2009),
631643.
[46] A. Nagalingam, M. Tighiouart, L. Ryden, L. Joseph, G. Landberg, N. K. Saxena, et al., Med1 plays a critical role in the
development of tamoxifen resistance, Carcinogenesis, 33 (2012),
918930.
[47] N. Navin, J. Kendall, J. Troge, P. Andrews, L. Rodgers,
J. McIndoo, et al., Tumour evolution inferred by single-cell
sequencing, Nature, 472 (2011), 9094.
[48] C. K. Ng, S. L. Cooke, K. Howe, S. Newman, J. Xian, J. Temple,
et al., The role of tandem duplicator phenotype in tumour
evolution in high-grade serous ovarian cancer, J Pathol, 226
(2012), 703712.
[49] D. G. Nord, ed., Intratumor Diversity and Clonal Evolution in
Cancer, vol. 112 of Advances in Cancer Research, Academic
Press, Oxford, 2011.
[50] P. C. Nowell, The clonal evolution of tumor cell populations,
Science, 194 (1976), 2328.
[51] S. Y. Park, M. Gonen, H. J. Kim, F. Michor, and K. Polyak,
Cellular and genetic diversity in the progression of in situ human
breast carcinomas to an invasive phenotype, J Clin Invest, 120
(2010), 636644.
[52] R. T. Pennock, Models, simulations, instantiations, and evidence:
the case of digital evolution, Journal of Experimental &
Theoretical Artificial Intelligence, 19 (2007), 2942.
[53] J. W. Pepper, Defeating pathogen drug resistance: guidance from
evolutionary theory, Evolution, 62 (2008), 31853191.
[54] J. W. Pepper, Somatic evolution of acquired drug resistance
in cancer, in Targeted Therapies: Mechanisms of Resistance,
D. Gioeli, ed., Springer-Verlag, New York, 2011, 127134.
[55] J. W. Pepper, Drugs that target pathogen public goods are robust
against evolved drug resistance, Evol Appl, 5 (2012), 757761.
[56] J. W. Pepper, C. S. Findlay, R. Kassen, S. L. Spencer, and C. C.
Maley, Cancer research meets evolutionary biology, Evol Appl,
2 (2009), 6270.
[57] O. Podlaha, M. Riester, S. De, and F. Michor, Evolution of the
cancer genome, Trends in Genetics, 28 (2012), 155163.
[58] G. Price, Selection and covariance, Nature, 227 (1970), 520521.
[59] D. F. Ransohoff, Promises and limitations of biomarkers, in
Cancer Prevention II, vol. 181 of Recent Results in Cancer
Research, Springer-Verlag, Berlin, 2009, 5559.
[60] S. Rosenfeld and I. Kapetanovic, Systems biology and cancer
prevention: all options on the table, Gene Regul Syst Bio, 2
(2008), 307319.
[61] S. Shah, R. Morin, J. Khattra, L. Prentice, T. Pugh, A. Burleigh,
et al., Mutational evolution in a lobular breast tumour profiled at
single nucleotide resolution, Nature, 461 (2009), 809813.

8
[62] S. Shah, A. Roth, R. Goya, A. Oloumi, G. Ha, Y. Zhao, et al.,
The clonal and mutational evolution spectrum of primary triplenegative breast cancers, Nature, 486 (2012), 395399.
[63] D. Shibata, Heterogeneity and tumor history, Science, 336
(2012), 304305.
[64] R. Siegel, D. Naishadham, and A. Jemal, Cancer statistics, 2012,
CA Cancer J Clin, 62 (2012), 1029.
[65] K. Sprouffske, J. W. Pepper, and C. C. Maley, Accurate
reconstruction of the temporal order of mutations in neoplastic
progression, Cancer Prev Res (Phila), 4 (2011), 11351144.
[66] R. F. Swaby and V. C. Jordan, Low-dose estrogen therapy to
reverse acquired antihormonal resistance in the treatment of
breast cancer, Clin Breast Cancer, 8 (2008), 124133.
[67] F. Thomas, D. Fisher, P. Fort, J. P. Marie, S. Daoust, B. Roche,
et al., Applying ecological and evolutionary theory to cancer: a
long and winding road, Evol Appl, 6 (2013), 110.
[68] V. G. Vogel, Selective estrogen receptor modulators and aromatase inhibitors for breast cancer chemoprevention, Curr Drug
Targets, 12 (2011), 18741887.
[69] V. G. Vogel, J. P. Costantino, D. L. Wickerham, W. M. Cronin,
R. S. Cecchini, J. N. Atkins, et al., Effects of tamoxifen vs
raloxifene on the risk of developing invasive breast cancer and
other disease outcomes: The NSABP Study of Tamoxifen and
Raloxifene (STAR) P-2 trial, JAMA, 295 (2006), 27272741.
Erratum in JAMA, 298 (2007), 973.
[70] M. J. Walter, D. Shen, L. Ding, J. Shao, D. C. Koboldt, K. Chen,
et al., Clonal architecture of secondary acute myeloid leukemia,
N Engl J Med, 366 (2012), 10901098.
[71] C. J. Ye, J. B. Stevens, G. Liu, S. W. Bremer, A. S. Jaiswal, K. J.
Ye, et al., Genome based cell population heterogeneity promotes
tumorigenicity: the evolutionary mechanism of cancer, J Cell
Physiol, 219 (2009), 288300.

Journal of Evolutionary Medicine